Academic Legal Writing - Eugene Volokh [183]
1. The First Claim
(i) The first error is small—the clause
a handgun is six times more likely to be used to kill a friend or relative than to repel a burglar
aims to summarize source A (“a handgun brought into the home for the purposes of self-protection is six times more likely to kill a relative or acquaintance than to repel a burglar”), but replaces “acquaintance” with “friend.” The two terms are false synonyms; they sound interchangeable, but they're different—members of rival gangs, for instance, may be acquaintances but not friends, and likewise for a drug dealer and his customer, or a prostitute and her client.* Not that huge a mistake, but worth avoiding all the same.
The more serious problem is that source A errs in quoting source B, which actually says
During the period surveyed in this study [1958—73 in Cuyahoga County, Ohio], only 23 burglars, robbers or intruders who were not relatives or acquaintances were killed by guns in the hands of persons who were protecting their homes. During this same interval, six times as many fatal firearm accidents occurred in the home.
Let's compare the student article and the original study (source B), noting the differences (tagged in italics):
(ii) The study discusses guns generally, not handguns in particular (item a). Nearly 2/3 of all guns in civilian hands are rifles or shotguns, not handguns.60Make clear when you're inferring from general data (which covers all guns) to a specific subset (handguns).
(iii) The cited study talks about fatal firearm accidents, which is not the same as killings of friends or relatives (item c). Most uses of firearms to kill a friend or relative are intentional killings, not accidents;61 and apparently about half of fatal firearm accidents involve the shooter accidentally killing himself, and others involve killings of strangers.62
(iv) “Repel[ling] a burglar” is different from “killing” one (item d). One can also repel a burglar with a handgun by visibly pointing it at him, by shooting and missing, or by shooting and wounding—and such uses are probably 50 or more times more common than killings of burglars.63Avoid false synonyms.
(v) “Burglar” is not the same as “burglars, robbers or intruders who were not relatives or acquaintances” (item e). The difference may not be great, but there is a difference.
(vi) The study is limited to one county and one period (item f). Gun crimes, accidents, and defensive uses vary by place and time; for instance, fatal gun accidents in the U.S. during the study period 1958–73 averaged about 2400/year, while in 2004–2006 they averaged about 700/year.64 It's therefore hard to tell how generalizable the study's findings are, but the article should certainly have acknowledged (at least in a footnote) that the study was limited to gun use in a particular place and time, and not to gun use generally. Make clear when you're inferring from a specific subset (Cuyahoga County, 1958–73) to general data (the country generally and at all times) or to another specific subset (the country in the year that the article was written).
The article said that “Proponents of manufacturers' liability further argue that handguns are almost useless for self-protection ....” Can the article's assertions be defended on the grounds that the article is only describing what proponents are arguing, not what is in fact true? If so, we shouldn't fault the author of the article being cite-checked, though we'd fault the author of source A.
But I don't think this is right. An author must expect that readers will interpret a statement like this as implicitly endorsing the cited statistics—and the article that we're checking has indeed been cited as endorsing the statistics that it describes. If authors want to cite erroneous sources only to show what others believe, they should explicitly state that the cited material is likely in error: This is part of their responsibility not to mislead their readers.
2. The Second Claim
The second claim is that “a person who uses a handgun in selfdefense is eight times more likely